Serilization and caching the table works!

.gitignore

@@ -12,3 +12,7 @@ kraklist.txt
 .*.un~
 papers
 callgrind*
+*.comp_new
+*.comp_bac
+bintest.bin
+*.dot

stdlib/grammer.krak

@@ -7,6 +7,7 @@ import symbol
 import regex
 import io
 import util
+import serialize
 
 fun split_into_words(gram_str: string::string): vector::vector<string::string> {
     var out.construct(): vector::vector<string>
@@ -100,7 +101,7 @@ fun load_grammer(gram_str: string::string): grammer {
     return gram
 }
 
-obj grammer (Object) {
+obj grammer (Object, Serializable) {
     var rules: vector::vector<rule>
     var non_terminals: set::set<symbol::symbol>
     var terminals: vector::vector<util::pair<symbol::symbol, regex::regex>>
@@ -133,6 +134,20 @@ obj grammer (Object) {
         parse_table.destruct()
     }
 
+    fun serialize(): vector::vector<char> {
+        return serialize::serialize(rules) + serialize::serialize(non_terminals) + serialize::serialize(terminals) + serialize::serialize(first_set_map) + serialize::serialize(parse_table)
+    }
+    fun unserialize(it: ref vector::vector<char>, pos: int): int {
+        // get everything constructed before the assignment
+        construct()
+        util::unpack(rules, pos) = serialize::unserialize<vector::vector<rule>>(it, pos)
+        util::unpack(non_terminals, pos) = serialize::unserialize<set::set<symbol::symbol>>(it, pos)
+        util::unpack(terminals, pos) = serialize::unserialize<vector::vector<util::pair<symbol::symbol, regex::regex>>>(it, pos)
+        util::unpack(first_set_map, pos) = serialize::unserialize<map::map<symbol::symbol, set::set<symbol::symbol>>>(it, pos)
+        util::unpack(parse_table, pos) = serialize::unserialize<table>(it, pos)
+        return pos
+    }
+
     fun calculate_first_set() {
         // the first set of a terminal is itself
         terminals.for_each( fun(terminal: util::pair<symbol::symbol, regex::regex>)
@@ -321,12 +336,30 @@ fun rule(lhs: symbol::symbol, rhs: vector::vector<symbol::symbol>): rule {
     return toRet
 }
 
-obj rule (Object) {
+obj rule (Object, Serializable) {
     var lhs: symbol::symbol
     var rhs: vector::vector<symbol::symbol>
     var position: int
     var lookahead: set::set<symbol::symbol>
 
+    fun serialize(): vector::vector<char> {
+        return serialize::serialize(lhs) + serialize::serialize(rhs) + serialize::serialize(position) + serialize::serialize(lookahead)
+    }
+    fun unserialize(it: ref vector::vector<char>, pos: int): int {
+        var tempLhs = symbol::invalid_symbol()
+        var tempRhs = vector::vector<symbol::symbol>()
+        var tempLookahead = set::set<symbol::symbol>()
+        util::unpack(tempLhs, pos) = serialize::unserialize<symbol::symbol>(it, pos)
+        util::unpack(tempRhs, pos) = serialize::unserialize<vector::vector<symbol::symbol>>(it, pos)
+        util::unpack(position, pos) = serialize::unserialize<int>(it, pos)
+        util::unpack(tempLookahead, pos) = serialize::unserialize<set::set<symbol::symbol>>(it, pos)
+
+        lhs.copy_construct(&tempLhs)
+        rhs.copy_construct(&tempRhs)
+        lookahead.copy_construct(&tempLookahead)
+        return pos
+    }
+
     fun construct(): *rule {
         lhs.construct()
         rhs.construct()
@@ -480,7 +513,7 @@ obj action {
     }
 }
 
-obj table (Object) {
+obj table (Object, Serializable) {
     // a 2 dimensional table made of a vector and a map that maps from stateno & symbol to a vector of parse actions
     var items: vector::vector<map::map<symbol::symbol, vector::vector<action>>>
 
@@ -497,6 +530,15 @@ obj table (Object) {
     fun destruct() {
         items.destruct()
     }
+    fun serialize(): vector::vector<char> {
+        return serialize::serialize(items)
+    }
+    fun unserialize(it: ref vector::vector<char>, pos: int): int {
+        var temp = vector::vector<map::map<symbol::symbol, vector::vector<action>>>()
+        util::unpack(temp, pos) = serialize::unserialize<vector::vector<map::map<symbol::symbol, vector::vector<action>>>>(it, pos)
+        items.copy_construct(&temp)
+        return pos
+    }
     fun expand_to(include_state: int) {
         while (include_state >= items.size)
             items.addEnd(map::map<symbol::symbol, vector::vector<action>>())

stdlib/io.krak

@@ -75,6 +75,22 @@ fun print(toPrint: double) : void{
 }
 
 // Ok, just some DEAD simple file io for now
+fun file_exists(path: string::string): bool {
+    var char_path = path.toCharArray()
+    defer delete(char_path)
+    var result = false
+	__if_comp__ __C__ {
+		simple_passthrough(char_path:result:) """
+            bool result = false;
+			FILE *fp = fopen(char_path, "r");
+            if (fp) {
+                result = true;
+                fclose(fp);
+            }
+		"""
+	}
+    return result
+}
 fun read_file(path: string::string): string::string {
     var toRet.construct(read_file_binary(path)): string::string
     return toRet

stdlib/map.krak

@@ -1,5 +1,7 @@
 import vector
 import io
+import serialize
+import util
 
 fun map<T,U>(): map<T,U> {
     var toRet.construct(): map<T,U>
@@ -11,13 +13,14 @@ fun map<T,U>(key:T, value:U): map<T,U> {
     return toRet
 }
 
-obj map<T,U> (Object) {
+obj map<T,U> (Object, Serializable) {
     var keys: vector::vector<T>
     var values: vector::vector<U>
 
-    fun construct() {
+    fun construct(): *map<T,U> {
         keys.construct()
         values.construct()
+        return this
     }
     fun copy_construct(old: *map<T,U>) {
         keys.copy_construct(&old->keys)
@@ -31,6 +34,18 @@ obj map<T,U> (Object) {
         keys.destruct()
         values.destruct()
     }
+    fun serialize(): vector::vector<char> {
+        return serialize::serialize(keys) + serialize::serialize(values)
+    }
+    fun unserialize(it: ref vector::vector<char>, pos: int): int {
+        var tempKeys = vector::vector<T>()
+        var tempValues = vector::vector<U>()
+        util::unpack(tempKeys, pos) = serialize::unserialize<vector::vector<T>>(it, pos)
+        util::unpack(tempValues, pos) = serialize::unserialize<vector::vector<U>>(it, pos)
+        keys.copy_construct(&tempKeys)
+        values.copy_construct(&tempValues)
+        return pos
+    }
     fun operator[]=(key: T, value: U) {
         set(key,value)
     }

stdlib/parser.krak

@@ -454,6 +454,15 @@ obj reduction (Object) {
     var nullable_parts: *tree<symbol>
     var label: *tree<symbol>
 
+    fun construct(): *reduction {
+        from = null<tree<int>>()
+        sym = invalid_symbol()
+        length = -1
+        nullable_parts = null<tree<symbol>>()
+        label = null<tree<symbol>>()
+        return this
+    }
+
     fun construct(f: *tree<int>, s: symbol, l: int, n: *tree<symbol>, labelIn:*tree<symbol>): *reduction {
         from = f
         sym.copy_construct(&s)

stdlib/regex.krak

@@ -5,6 +5,7 @@ import mem
 import set
 import util
 import conversions
+import serialize
 
 fun regex(in: *char):regex {
     return regex(string::string(in))
@@ -40,7 +41,7 @@ obj regexState (Object) {
     }
 }
 
-obj regex (Object) {
+obj regex (Object, Serializable) {
     var regexString: string::string
     var begin: *regexState
     var referenceCounter: *int
@@ -85,6 +86,15 @@ obj regex (Object) {
             mem::delete(referenceCounter)
         }
     }
+    fun serialize(): vector::vector<char> {
+        return serialize::serialize(regexString)
+    }
+    fun unserialize(it: ref vector::vector<char>, pos: int): int {
+        var temp = string::string()
+        util::unpack(temp, pos) = serialize::unserialize<string::string>(it, pos)
+        construct(temp)
+        return pos
+    }
 
     fun operator==(other: regex):bool {
         return regexString == other.regexString

stdlib/serialize.krak

@@ -16,13 +16,13 @@ fun serialize<T>(it: T): vector::vector<char> {
 }
 
 // dead simple wrapper for ease of use
-fun unserialize<T>(it: vector::vector<char>): T {
+fun unserialize<T>(it: ref vector::vector<char>): T {
     return unserialize<T>(it, 0).first
 }
-fun unserialize<T>(it: vector::vector<char>, pos: int): util::pair<T,int> {
+fun unserialize<T>(it: ref vector::vector<char>, pos: int): util::pair<T,int> {
     return util::make_pair(*conversions::cast_ptr<*char,*T>(it.getBackingMemory()+pos), pos + mem::sizeof<T>())
 }
-fun unserialize<T(Serializable)>(it: vector::vector<char>, pos: int): util::pair<T,int> {
+fun unserialize<T(Serializable)>(it: ref vector::vector<char>, pos: int): util::pair<T,int> {
     var toRet: T
     pos = toRet.unserialize(it, pos)
     return util::make_pair(toRet, pos)

stdlib/set.krak

@@ -1,5 +1,7 @@
 import vector
 import io
+import serialize
+import util
 
 fun set<T>(): set<T> {
     var toRet.construct() : set<T>
@@ -18,10 +20,11 @@ fun from_vector<T>(items: vector::vector<T>): set<T> {
     return toRet
 }
 
-obj set<T> (Object) {
+obj set<T> (Object, Serializable) {
     var data: vector::vector<T>
-    fun construct() {
+    fun construct(): *set<T> {
         data.construct()
+        return this
     }
     fun copy_construct(old: *set<T>) {
         data.copy_construct(&old->data)
@@ -30,6 +33,15 @@ obj set<T> (Object) {
         destruct()
         copy_construct(&rhs)
     }
+    fun serialize(): vector::vector<char> {
+        return serialize::serialize(data)
+    }
+    fun unserialize(it: ref vector::vector<char>, pos: int): int {
+        var temp = vector::vector<T>()
+        util::unpack(temp, pos) = serialize::unserialize<vector::vector<T>>(it, pos)
+        data.copy_construct(&temp)
+        return pos
+    }
     fun operator==(rhs: set<T>): bool {
         if (size() != rhs.size())
             return false

stdlib/stack.krak

@@ -1,4 +1,6 @@
 import vector
+import serialize
+import util
 
 
 fun stack<T>():stack<T> {
@@ -11,7 +13,7 @@ fun stack<T>(in:T):stack<T> {
     return out
 }
 
-obj stack<T> (Object) {
+obj stack<T> (Object, Serializable) {
     var data: vector::vector<T>
     fun construct(): *stack<T> {
         data.construct()
@@ -26,6 +28,15 @@ obj stack<T> (Object) {
     fun operator=(other: ref stack<T>) {
         data = other.data
     }
+    fun serialize(): vector::vector<char> {
+        return serialize::serialize(data)
+    }
+    fun unserialize(it: ref vector::vector<char>, pos: int): int {
+        var temp = vector::vector<T>()
+        util::unpack(temp, pos) = serialize::unserialize<vector::vector<T>>(it, pos)
+        data.copy_construct(&temp)
+        return pos
+    }
     fun push(it: ref T) {
         data.addEnd(it)
     }

stdlib/string.krak

@@ -1,6 +1,7 @@
 import vector
 import util
 import mem
+import serialize
 
 fun to_string(in: int): string {
     var dest = mem::new<char>(mem::sizeof<int>() * 8)
@@ -23,8 +24,11 @@ fun string(in:char):string {
     out += in
     return out
 }
+fun string():string {
+    return string("")
+}
 
-obj string (Object) {
+obj string (Object, Serializable) {
     var data: vector::vector<char>;
     fun construct(): *string {
         data.construct();
@@ -69,6 +73,15 @@ obj string (Object) {
         data.destruct()
     }
 
+    fun serialize(): vector::vector<char> {
+        return serialize::serialize(data)
+    }
+    fun unserialize(it: ref vector::vector<char>, pos: int): int {
+        construct()
+        util::unpack(data, pos) = serialize::unserialize<vector::vector<char>>(it, pos)
+        return pos
+    }
+
     fun operator[](index: int): ref char { return data[index]; }
     fun slice(first: int, second: int): string {
         var new.construct(data.slice(first,second)): string

stdlib/symbol.krak

@@ -1,4 +1,7 @@
 import string
+import serialize
+import vector
+import util
 
 fun null_symbol(): symbol {
     var toRet.construct(string::string("$NULL$"), false, string::string("$NULL$")): symbol
@@ -33,7 +36,7 @@ fun symbol(nameIn: string::string, terminalIn: bool, dataIn: string::string): sy
     return toRet
 }
 
-obj symbol (Object) {
+obj symbol (Object, Serializable) {
     var data: string::string
     var name: string::string
     var terminal: bool
@@ -62,6 +65,19 @@ obj symbol (Object) {
         destruct()
         copy_construct(&old)
     }
+    fun serialize(): vector::vector<char> {
+        return serialize::serialize(data) + serialize::serialize(name) + serialize::serialize(terminal)
+    }
+    fun unserialize(it: ref vector::vector<char>, pos: int): int {
+        var tempData = string::string()
+        var tempName = string::string()
+        util::unpack(tempData, pos) = serialize::unserialize<string::string>(it, pos)
+        util::unpack(tempName, pos) = serialize::unserialize<string::string>(it, pos)
+        util::unpack(terminal, pos) = serialize::unserialize<bool>(it, pos)
+        data.copy_construct(&tempData)
+        name.copy_construct(&tempName)
+        return pos
+    }
     fun operator==(other: ref symbol): bool {
         return data == other.data && name == other.name && terminal == other.terminal;
     }

stdlib/util.krak

@@ -1,5 +1,6 @@
 import mem
 import vector
+import serialize
 
 fun max<T>(a: T, b: T): T {
     if (a > b)
@@ -35,7 +36,7 @@ obj unpack_dummy<T,U> {
 }
 
 
-obj pair<T,U> (Object) {
+obj pair<T,U> (Object, Serializable) {
     var first: T
     var second: U
 
@@ -60,6 +61,17 @@ obj pair<T,U> (Object) {
         mem::maybe_destruct(&first)
         mem::maybe_destruct(&second)
     }
+    fun serialize(): vector::vector<char> {
+        return serialize::serialize(first) + serialize::serialize(second)
+    }
+    fun unserialize(it: ref vector::vector<char>, pos: int): int {
+        // can't use unpack :( (b/c we can't make an already constructed empty one)
+        var first_pair = serialize::unserialize<T>(it, pos)
+        var second_pair = serialize::unserialize<U>(it, first_pair.second)
+        mem::maybe_copy_construct(&first, &first_pair.first)
+        mem::maybe_copy_construct(&second, &second_pair.first)
+        return second_pair.second
+    }
 
     // the old unnecessary template to prevent generation
     // if not used trick (in this case, changing out U with V)

stdlib/vector.krak

@@ -46,7 +46,7 @@ obj vector<T> (Object, Serializable) {
             toRet += serialize(data[i])
         return toRet
     }
-    fun unserialize(it: vector<char>, pos: int): int {
+    fun unserialize(it: ref vector<char>, pos: int): int {
         unpack(size, pos) = unserialize<int>(it, pos)
         data = new<T>(size)
         available = size

tests/test_grammer.krak

@@ -6,17 +6,81 @@ import string:*
 import util:*
 import symbol:*
 import tree:*
+import serialize:*
 
 fun main():int {
     
-    var a = load_grammer(read_file(string("../krakenGrammer.kgm")))
+    var a.construct(): grammer
-    /*var a = load_grammer(read_file(string("grammer.kgm")))*/
+
-    /*var a = load_grammer(read_file(string("grammer2.kgm")))*/
+    var file_name = string("../krakenGrammer.kgm")
-    /*var a = load_grammer(read_file(string("grammer3.kgm")))*/
+    /*var file_name = string("grammer.kgm")*/
-    /*var a = load_grammer(read_file(string("grammer4.kgm")))*/
+    /*var file_name = string("grammer2.kgm")*/
+    /*var file_name = string("grammer3.kgm")*/
+    /*var file_name = string("grammer4.kgm")*/
+
+    var compiled_name = file_name + string(".comp_new")
+    var file_contents = read_file(file_name)
+    var loaded_and_valid = false
+
+/*
+    println("gonna serialize")
+    var s = serialize(file_contents)
+    println("gonna write")
+    write_file_binary(compiled_name, s)
+    println("gonna read")
+    var bin = read_file_binary(compiled_name)
+    println("gonna setup")
+    var pos = 0
+    var uns = string()
+    println("gonna unserialize")
+    unpack(uns, pos) = unserialize<string>(bin, pos)
+    println("gonna done")
+
+    return 0
+    */
+
+        /*a = load_grammer(file_contents)*/
+        /*println("grammer loaded, calculate_state_automaton")*/
+        /*a.calculate_first_set()*/
+        /*a.calculate_state_automaton()*/
+    if (file_exists(compiled_name)) {
+        println("cached file exists")
+        var pos = 0
+        var binary = read_file_binary(compiled_name)
+        println("read file!")
+        var cached_contents = string()
+        println("made tmp string!")
+        unpack(cached_contents, pos) = unserialize<string>(binary, pos)
+        println("unserialized the string!")
+        if (cached_contents == file_contents) {
+            println("loaded_and_valid, using cached version!")
+            loaded_and_valid = true
+            unpack(a, pos) = unserialize<grammer>(binary, pos)
+            println("finished unserializeing!!")
+        } else {
+            println("file contents do not match:")
+            println("CACHED:")
+            println(cached_contents)
+            println("REAL:")
+            println(file_contents)
+            println("END")
+        }
+    } else {
+        println("cached file does not exist")
+    }
+    if (!loaded_and_valid) {
+        println("Not loaded_and_valid, re-generating and writing out")
+        a = load_grammer(file_contents)
+        println("grammer loaded, calculate_first_set")
+        a.calculate_first_set()
+        println("grammer loaded, calculate_state_automaton")
+        a.calculate_state_automaton()
+        println("calculated, writing out")
+        write_file_binary(compiled_name, serialize(file_contents) + serialize(a))
+        println("done writing")
+    }
     println(a.to_string())
     var doFirstSet = fun() {
-        a.calculate_first_set()
         println("///////////////////START FIRST SET/////////////")
         println("//TERMINALS//")
         a.terminals.for_each( fun(terminal: util::pair<symbol::symbol, regex::regex>) {
@@ -39,7 +103,7 @@ fun main():int {
         })
         println("///////////////////END FIRST SET/////////////")
     }
-    doFirstSet()
+    /*doFirstSet()*/
 
     var lex = lexer(a.terminals)
 
@@ -50,8 +114,9 @@ fun main():int {
     println("woo lexing:")
     /*range(8).for_each(fun(i: int) { println(lex.next().to_string()); } )*/
     /*range(80).for_each(fun(i: int) { println(lex.next().to_string()); } )*/
-    println(a.to_string())
+    /*println(a.to_string())*/
-    a.calculate_state_automaton()
+
+
     var parse.construct(a): parser
     /*var result = parse.parse_input(string("a"), string("fun name"))*/
     var result = parse.parse_input(read_file(string("to_parse.krak")), string("fun name"))

tests/test_serialization.expected_results

@@ -1,3 +1,6 @@
 7 = 7
 9 , 11 = 9 , 11
 1 2 3 4 5 
+hello serialize
+3 = 2.700000
+50 = 3.141590

tests/test_serialization.krak

@@ -4,6 +4,7 @@ import string:*
 import util:*
 import vector:*
 import vector_literals:*
+import map:*
 
 fun main():int {
     var intA = 7
@@ -43,6 +44,28 @@ fun main():int {
     back.for_each(fun(i: int) { print(i); print(" "); })
     println()
 
+    // ok, lets do a string
+    write_file_binary(string("bintest.bin"), serialize(string("hello serialize")))
+    bin = read_file_binary(string("bintest.bin"))
+    var backStr = string()
+    pos = 0
+    unpack(backStr, pos) = unserialize<string>(bin, 0)
+    println(backStr)
+
+    // ok, lets do a map
+    var m = map(3, 2.7)
+    m.set(50, 3.14159)
+    write_file_binary(string("bintest.bin"), serialize(m))
+    bin = read_file_binary(string("bintest.bin"))
+    var backM = map<int,double>()
+    pos = 0
+    unpack(backM, pos) = unserialize<map<int,double>>(bin, 0)
+    backM.for_each(fun(key: int, value: double) {
+        print(key)
+        print(" = ")
+        println(value)
+    })
+
     return 0
 }